Geochimica et Cosmochimica Acta, Vol. 68, No. 17, pp. 3497–3507, 2004 Copyright © 2004 Elsevier Ltd Pergamon Printed in the USA. All rights reserved 0016-7037/04 $30.00 ϩ .00 doi:10.1016/j.gca.2004.02.019 The oxidation of carbonate green into ferric phases:solid-state reaction or transformation via solution

1, 2 1 LUDOVIC LEGRAND, *LEO´ MAZEROLLES, and ANNIE CHAUSSE´ 1Laboratoire Analyse et Environnement, UMR 8587, CEA-CNRS-Universite´ d’Evry Val d’Essonne, Rue du pe`re Jarland, F-91025 Evry France 2Centre d’Etudes de Chimie Me´tallurgique, CNRS, F-94407 Vitry sur Seine France

(Received February 28, 2003; accepted in revised form February 18, 2004)

2Ϫ ϭ Abstract—The oxidation of carbonate green rust, GR(CO3 ), in NaHCO3 solutions at T 25°C has been 2Ϫ investigated through electrochemical techniques, FTIR, XRD, TEM and SEM. The used GR(CO3 ) samples were made of either suspended solid in solution or a thin electrochemically formed layer on the surface of an disc. Depending on experimental conditions, oxidation occurs, with or without major modifications of the 2Ϫ GR(CO3 ) structure, suggesting the existence of two pathways: solid-state oxidation (SSO) leading to a ferric oxyhydroxycarbonate as the end product, and a dissolution-oxidation-precipitation (DOP) mechanism leading to ferric oxihydroxides such as , , or . A formula was proposed for this ferric III oxyhydroxycarbonate, Fe6 O(2ϩx)(OH)(12-2x)(H2O)x(CO3), assuming that the solid-state oxidation reaction is associated to a deprotonation of the water molecules within the interlayers, or of the hydroxyl groups in the Fe(O,H) octahedra layers. The DOP mechanism involves transformation via solution with the occurrence of soluble ferrous-ferric intermediate species. A discussion about factors influencing the oxidation of carbonate 2Ϫ green rust is provided hereafter. The ferric oxyhydroxycarbonate can be reduced back to GR(CO3 )bya reverse solid-state reduction reaction. The potentiality for a solid-state cycling of iron to occur may be considered. The stability of the ferric oxyhydroxycarbonate towards thermodynamically stable ferric phases, such as goethite and hematite, was also studied. Copyright © 2004 Elsevier Ltd

1. INTRODUCTION very improbable, on the basis of FTIR analysis and site avail- ability in interlayers (Benali et al., 2001; Legrand et al., 2001a). Green (GRs) are layered Fe(II)-Fe(III) hydroxi-salts Synthetic GRs have been prepared in the laboratory, using that are suspected to occur as minerals in soils alternating redox various chemical procedures, from solutions containing ferrous conditions (Ponnamperuma, 1972; Lindsay, 1979; Taylor, species and chloride (Feitknecht and Keller, 1950; Detour- 1980; Trolard et al., 1997). The bluish-green colour of such nay et al., 1976; Refait and Ge´nin, 1993; Schwertmann and soils, which turns ochre once exposed to air, would be due to Fechter, 1994), ferrous species and sulphate ions (Detournay et the presence of these mixed-valence compounds. This natural al., 1975; Olowe and Ge´nin, 1991; Ge´nin et al., 1996), and occurrence was postulated on the basis of Mo¨ssbauer spectros- ferrous species and carbonate ions (Taylor, 1980; Hansen, copy analysis. Green rusts have also been identified in aerobic/ 1989; Drissi et al., 1995). Carbonate green rust films have also anaerobic (Butler and Beynon, 1967; Stampfl, 1969; been obtained from the electrochemical oxidation of iron discs McGill et al., 1976; Ge´nin et al., 1993; Abdelmoula et al., (Legrand et al., 2000). 1996; Bonin et al., 2000; Savoye et al., 2001). Recent studies have reported the formation of green rusts Green rusts belong to the general class of layered double from the bioreduction of ferric oxihydroxides by dissimilatory- (LDH) characterised by a crystalline structure con- iron-reducing bacteria (DIRB) (Fredrickson et al., 1998; Ona- sisting in the stacking of brucite-like layers carrying positive Nguema et al., 2002). Such bacteria can use the ferric oxihy- charges and interlayers constituted by anions and water mole- droxide as an electron acceptor for the oxidation of organic cules. The structure and the chemical formula of GRs depend matter. Ona-Nguema et al. (2002) have shown that large hex- upon the specific anions AnϪ that they incorporate (AnϪ agonal crystals of carbonate green rust could be obtained, and ϭ Ϫ 2Ϫ 2Ϫ they stated that this process may occur in soil solutions and Cl ,SO4 ,CO3 ...) (Bernal et al., 1959; Ge´nin et al., 2Ϫ aquifers. 1998). Carbonate green rust, GR(CO3 ), has a trigonal struc- ture looking like that of pyroaurite with a stacking sequence The chemistry and mobility of iron is strongly related to the AcBiBaCjCbAkA..., where A, B, and C represent OHϪ planes, redox conditions. In environments alternating oxidising and a,b, and c Fe(II)-Fe(III) cations layers and i,j, and k intercalated reducing conditions, the cycling of iron will induce oxidation of layers (Allmann, 1968). The Fe(II)/Fe(III) ratio usually found green rust by dissolved oxygen into ferric products. This oxi- for carbonate green rust is (1). (Hansen, 1989) and the chemical dation process also occurs when GR interacts with oxidising Ϫ Ϫ II III 2ϩ 2Ϫ 2Ϫ species such as NO (Hansen and Bender Koch, 1998), CrO2 formula, [Fe4 Fe2 (OH)12] . [CO3 .m H2O] , should be pro- 3 4 posed by analogy with pyroaurite. Previous studies have shown (Loyaux-Lawniczak et al., 2000; Williams and Scherer, 2001), Ϫ Ϫ 2Ϫ CCl (Erbs et al., 1999), SeO2 (Refait et al., 2000), both in that incorporation of HCO3 within the lattice of GR(CO3 )is 4 4 natural aquifers and iron permeable reactive barriers (PRBs). Several studies were dedicated to the characterisation of the * Author to whom correspondence should be addressed products resulting from the oxidation of carbonate green rust. ([email protected]). However, the reaction pathways have not been fully under- 3497 3498 L. Legrand, L. Mazerolles, and A. Chausse´

stood. Drissi et al. (1995) suggested that the oxidation of electrolysis, and the argon bubbling was maintained during the whole 2Ϫ Ϫ 2Ϫ electrochemical measurements. GR(CO3 )inHCO3 /CO3 solutions leads first to a so-called “amorphous active FeOOH” phase distinct from ␣ or Electrochemical preparation and investigations were performed us- ␥ ing a potentiostat PAR 273 A controlled by an IBM computer and the -FeOOH. In a recent study, Benali et al. (2001) asserted that PAR Model 270 software package. The electrochemical synthesis of 2Ϫ Ϫ this intermediate phase is ferrihydrite, and proposed the fol- GR(CO3 ) layer was done by applying an anodic potential ( 0.45 2Ϫ V ) to the iron disk in 0.4 mol/L NaHCO solution at pH ϭ 9.6. The lowing sequence for the oxidation: GR(CO3 ) oxidizes into SHE 3 ferrihydrite, and then, ferrihydrite transforms into goethite, electrolysis was stopped as the current density became lower than 10 ␮AcmϪ2, indicating that the electrode surface was fully covered by the ␣-FeOOH, via a dissolution-precipitation mechanism. Taylor 2Ϫ GR(CO3 ) particles. The coulombic charge required for this purpose (1980), Hansen (1989) and Abdelmoula et al. (1996) studied was ϳ150 mC cmϪ2. More details about the electrochemical synthesis 2Ϫ the oxidation by air of dry samples of carbonate green rust. The and FTIR analysis of GR(CO3 ) layers can be found in Legrand et al. formation of feroxyhite, ␦-FeOOH, or ferrihydrite was claimed, (2001a and 2001b). based on the presence of a line at 0.254 nm in the diffraction pattern. And last, a study by Markov et al. (1990) in which an 2.3. Characterisation of Solid Products iron(III) hydroxide carbonate was obtained from the decompo- XRD measurements were carried out by using a Philips PW1830 sition of siderite in wet air, can also be mentioned. diffractometer with a Co K␣ radiation (1.7902 A˚ ). The samples were The aim of the present work is to provide a better description scanned from 5° to 90° 2␪. FTIR data were recorded on a Bruker IFS of the oxidation process of carbonate green rust and the result- 28 Fourier Transform Infrared spectrometer. Powder samples were pressed to pellets with KBr and analysed by direct transmission mode. ing ferric products. The transformation of carbonate green rust Electrochemically formed layers on the surface of the iron disk were under various oxidising conditions was studied through elec- directly analysed using a reflection transmission tool (Grazeby- trochemical measurements, X-ray diffraction, FTIR spectros- Specac). 20 scans were done for the spectrum acquisition and the copy, TEM, and SEM. acquisition time was ϳ30 s. TEM measurements were performed with a JEOL 2000FX operating at 200kV. A double-tilt Ϯ30° specimen holder was used, and observations were directly achieved on powders 2. EXPERIMENTAL METHODS deposited on copper grids. SEM measurements were performed with a .Synthesis and Oxidation of GR(CO2؊) Suspension Philips XL30 microscope .2.1 3 Solid products, resulting from chemical synthesis, were separated ␮ 50 mL of 0.4 or 0.04 mol/L NaHCO3 solution (prepared with from the solutions by collecting on 0.22 m Millipore filters under Ͼ ⍀ NaHCO3 as received from Fluka, purity 99.5%, and 18 M cm argon atmosphere. They were rinsed with deaerated nano-pure water. nano-pure water) were introduced in a glass cell (ϳ50 mm diam.) with The filters were then quickly removed from the filtration tool and five holes on the top. Three of the holes were used for the electrodes: introduced into a closed glass tube equipped with argon inlet and outlet

platinum electrode, saturated calomel electrode (0.25 VSHE), and com- for drying. The solid products were maintained under argon flow for bined pH electrode (WTW sentix 41). Another hole was used for the ϳ1 h before analysis. In the case of the intermediate product (carbonate argon inlet and outlet. The fifth hole 15 mm diameter), located in the green rust), the preparation of the KBr pellets for FTIR measurements middle position, was closed by a rubber cap during the deaeration of the was done under argon atmosphere to avoid oxidation by air. For XRD solution; for the aeration of the solution, the argon flow was stopped measurements, the intermediate product was protected from air oxida- and the rubber cap was removed to allow the air oxygen to get in tion by admixing glycerol, once filtered (Hansen, 1989). Oxidation contact with the solution at the air/liquid interface. The solution was products electrochemically formed at the surface of an iron disk were magnetically stirred (300 rpm) and thermostated at 25°C during the washed with deaerated 18 M⍀ cm nano-pure water, dried under argon chemical procedure. The solution/air interface was ϳ20 cm2 in area. flow, and quickly introduced into the chamber for FTIR or SEM

Before Fe(II) addition, the NaHCO3 solution was deaerated with argon, analysis. The contact with air was less than 1 min. (argon U, Air liquide, ϳ20 mL minϪ1) whose residual oxygen content The following procedure was used to determine the Fe(II) content in

was eliminated by bubbling in a Fe(OH)2 suspension present in the solid suspensions at a given time. After removal of air by argon purge liner before injection into the cell. Then 0.5 mL (giving an initial Fe(II) to stop the oxidation, the solution was acidified with concentrated 2Ϫ ϳ concentration of 10 M) of a 1 mol/L FeCl2 solution was injected into H2SO4 down to a pH of 0.5 to induce the dissolution of the solid 2ϩ the solution using a syringe through the rubber cap. To get an appro- phases. Titration of Fe ions was done with a 0.01 mol/L KMnO4 priate initial pH value, 10 mol/L (or 1 mol/L) NaOH solution was solution.

dropped into the preparation. FeCl2 solution was prepared by dissolv- The total iron content was assessed by two methods. The first one ing the appropriate amount of FeCl2,4H2O from Aldrich (purity 99%) consists in a complexometric titration of Fe(III) by ethylene diamine in deaerated 18 M⍀ cm nano-pure water. The oxidation reaction was tetraacetic acid (EDTA from Prolabo, normapur). 50 mg of dried solids ϳ monitored by recording both redox potential (Eh) and pH of the (50 mg) are dissolved in hot 1 mol/L HCl ( 5–7 mL). After complete solution. A millivoltmeter (Taccusel minisis 8000) and a pHmeter dissolution, 18 M⍀ cm nano-pure water is added until a volume of 50 (WTW multiline P4) were used. In some experiments, hydrogen per- mL si reached. The solution is then thermostated at 45°C, and 3 drops oxide was also used as an oxidising species. A 1 mol/L solution was of sulfosalicilic acid indicator are added. The equivalent volume of

prepared with 30% H2O2 commercial solution (as received from EDTA (0.05 mol/L) is measured when the colour changes from dark ⍀ 2Ϫ Sigma) and 18 M cm nano-pure water. purple to light yellow. In the case of GR(CO3 ), 2 mmol of H2O2 are added to the solution before titration, to oxidise Fe2ϩ ions into Fe3ϩ Electrochemical Studies of GR(CO2؊) Thin Layer On Iron ions. The second method consists in hematisation (i.e., transformation .2.2 3 ␣ of the product into hematite, -Fe2O3). The dried product is first Electrochemical measurements were performed in a cylindrical cell weighted (m0) and then heated to 300°C for 40 h. After this treatment, containing ϳ15 mL of solution. A 15 mm diameter iron disk (99.5% the product is fully transformed into hematite, whose mass can be

purity, provided by Goodfellow) was used as a working electrode; a determined (mH). The mass of iron, mFe, is equal to 0.699mH. Finally, large surface-area platinum grit was used as a counter electrode; an the total Fe content in the solid product is determined by the mFe to m0 AgCl-coated silver wire immersed in a compartment with a fine po- ratio. It has to be noted that the complete transformation into hematite rosity glass frit and filled with 0.1 mol/L NaCl was used as a reference is controlled by FTIR.

electrode (0.28 VSHE). The iron disk was polished on abrasive foils The CO2 content in the solid samples was determined by acid-base from grade 500 to grade 4000 and rinsed with 18 M⍀ cm nano-pure titration. A weighted amount of dried solid sample (in the 52–57 mg water before use. The iron disk was then placed at the bottom of the range) was introduced into an inner glass tube (ϳ10 mL) located in an cell. The surface in contact with the electrolyte was measured as 0.785 outer glass tube (ϳ50 mL) filled in with 5 mL of 0.1 mol/L NaOH cm2. Electrolyte solutions were deaerated with argon for 15 min before solution; the latter solution was prepared from 30% NaOH solution (RP Oxidation of carbonate green rust into ferric phases 3499

3.1.2. Characterisation of solid products Figure 2 gives the XRD patterns of the products obtained at points A, B, and C. The initial ferrous suspension (a) is amor- phous as shown by the lack of diffraction lines. Figure 2b exhibits typical lines of a pyroaurite-like green rust structure (Allmann, 1968). Following lattice parameter values were cal- culated assuming a rhombohedral R3៮m structure,

a ϭ 0.3172 Ϯ 0.0005 nm

and

c ϭ 2.274 Ϯ 0.0 3 nm.

They are consistent with those reported in the literature for 2Ϫ carbonate green rust GR(CO3 ) (McGill et al., 1976; Taylor, 1980; Hansen, 1989; Drissi et al., 1995; Legrand et al., 2001a). 2Ϫ ៮ As GR(CO3 ) belongs to the R3m group, c parameter indicates a 0.758 nm interplanar distance between Fe2ϩ-Fe3ϩ layers. Fig. 1. E vs. time and %Fe(II) vs. time recorded during oxidation by XRD pattern of the ferric product, sampled at point C (Fig. 2c), Ϫ2 ϭ displays a main line, corresponding to an interplanar distance of air of 10 M Fe(II) suspension in 0.4 mol/L NaHCO3 solution; pH 9.6; T ϭ 25°C; stirring 300 rpm. 0.734 nm. Some minor lines at 0.367 nm (0.734/2), 0.253 nm, and 0.226 nm are also observed. Such a diagram has already 2Ϫ been reported in the literature for GR(CO3 ) powders oxidised by air in dry conditions (Taylor, 1980; Hansen, 1989; Abdel- Prolabo) containing less than 1% of Na2CO3 impurity. Both tubes were equipped with a magnetic bar. The inner tube was maintained open moula et al., 1996). The 0.734 nm interplanar distance value while the outer one was closed by a gas-proof rubber stopper. Then, 7 may indicate that some relics of the lamellar structure of the mL of 1 mol/L HCl solution were injected into the inner tube through 2Ϫ parent GR(CO3 ) remain in the Fe(III) product structure ob- the stopper by using a syringe. The tubes were settled on a magnetic 2ϩ 3ϩ stirrer (300 rpm) for 5 h. The suspended solid completely dissolved tained at point C. Since the ionic radii of Fe and Fe are within 3 h. CO2 resulted from the dissolution and was absorbed in 0.078 and 0.0645 nm, respectively, the difference between NaOH solution. CO2 was then determined by acid/base titration of the 0.734 nm and 0.758 nm could be related to a contraction in the carbonated NaOH solution. Note that the blank titration was carried out and the CO2 content of the 0.1 mol/L NaOH solution was subtracted from the experimental values.

3. RESULTS

2؊ 3.1. Synthesis and Oxidation of GR(CO3 ) Suspension In 0.4 mol/L NaHCO3 Solution

3.1.1. E-t and pH-t transients

Figure 1 reports the potential values recorded with a plati- num electrode as a function of time during the aerial oxidation of a Fe(II) suspension in 0.4 mol/L NaHCO3 solution with an initial pH ϭ 9.6, and the corresponding evolution of Fe(II) content, expressed as a percentage. Oxidation occurs in two steps separated by a potential jump. The potential jump indi- cates the end of the first step, corresponding to the complete consumption of the initial ferrous suspension. Analysis done at point B indicates that the Fe(II) content is ϳ66%. The colour of the suspension changes from white to dark green (from A to B), then to greenish-brown, and finally to light brown colours (from B to C). From A to C, the pH does not significantly change. No potential jump is observed at the end of the second step, but Fe(II) titration indicates that the formation of the ferric product gets completed at point C (0% of Fe(II)). The average oxidation rates for the A-to-B and B-to-C steps are 6.0 and 2.3 II Ϫ1 Fig. 2. XRD patterns of (a) the starting ferrous suspension sampled ␮moles Fe min , respectively. Experiments performed at 2Ϫ at point A; (b) the product sampled at point B, GR(CO3 ); and (c) the different initial pH values in the 8.8 to 10.5 range gave E-t ferric product sampled at point C, exGR-Fe(III). (* goethite as a trace). transients similar to that reported in Figure 1. Step time (a and b) 3 s; (c) 5 s; step size 2␪ (a and b) 0.025°; (c) 0.03°. 3500 L. Legrand, L. Mazerolles, and A. Chausse´

3020 cmϪ1. The absorption at 1635 cmϪ1 is assigned to the 2Ϫ bending vibration of water. The presence of CO3 ions is ␯ revealed by the two strong absorption bands at 1360 ( 3) and Ϫ1 ␯ ␯ 690 cm ( 4) and two much weaker ones at 1070 ( 1) and 850 Ϫ1 ␯ 2Ϫ cm ( 2). The symmetry of CO3 deviates from D3h as seen ␯ Ϫ1 from the splitting of the 3 mode (1360 and 1480 cm ) and ␯ from the observation of the normally IR-forbidden 1 absorp- 2Ϫ tion. To prove that CO3 anions are inside the lattice rather than adsorbed species, an exGR-Fe(III) sample was synthesized according to the same procedure as given in Figure 1, except Ϫ2 that 10 MNa2HPO4 was added at point B. The final product is also exGR-Fe(III), as revealed by XRD (pattern not shown) and FTIR analysis (Fig. 3d). The adsorption of phosphate ␯ species is indicated by the antisymmetric 3 PO4 stretching Ϫ1 ␯ (1000 and 1105 cm ) and 4 PO4 deformation (620 and 640 cmϪ1) modes, which split due to the monodentate or bidentate coordination to the oxo or hydroxo-groups. The carbonate bands are not affected by the presence of adsorbed phosphate species. The transmission electron microscopy image of exGR- Fe(III) is given in Figure 4. It shows a hexagonal platelets morphology identical to that commonly reported in the litera- ture for green rust (McGill et al., 1976; Legrand et al., 2001a; Ge´hin et al., 2002; Peulon et al., 2003), suggesting a solid-state oxidation of carbonate green rust into exGR-Fe(III). The elec- tron diffraction studies exploring the reciprocal lattice did not allow us to identify the structure of this phase due to the very Fig. 3. FTIR spectra of (a) the starting ferrous suspension sampled at anisotropic morphology of these platelets. This anisotropy lim- 2Ϫ point A; (b) the product sampled at point B, GR(CO3 ); (c) the ferric its the variety of cristallographic planes which can be studied, product sampled at point C, exGR-Fe(III); and (d) the ferric product and it is not possible to measure the c parameter that is sampled at point C, following the same procedure as given in Figure 1, Ϫ2 perpendicular to the observation plane. except that 10 MNa2HPO4 was added at point B. 2Ϫ Table 1 reports the results of chemical analysis of GR(CO3 ) and exGR-Fe(III) samples and the data computed for some ferric oxihydroxides that should be expected according to the solid lattice induced by the Fe2ϩ-to-Fe3ϩ transformation (Ab- literature (see Introduction). The total Fe and CO2 contents of delmoula et al., 1996). Moreover, the small variation of the GR(CO2Ϫ) and exGR-Fe(III) compare quite well. The theoret- interplanar distance might be consistent with the presence of 3 2Ϫ ical values calculated for ferrihydrite and feroxyhyte do not CO3 anions in the interlayers of the Fe(III) product. The ferric match at all with the experimental data obtained for exGR- product sampled at point C will thereafter be referred as exGR- Fe(III). Fe(III). The FTIR spectrum of the initial ferrous suspension (Fig. 3a) exhibits bands at 1530, 1355, 1075, 955, 840, 780, and 760 cmϪ1, that have already been reported for ferrous hydroxicar-

bonate, Fe2(OH)2CO3, (Erdo¨s and Altorfer, 1976). The broad- ness of the IR bands is consistent with the amorphous nature of the solid product, as pointed out by XRD. The FTIR spectrum of the product formed at point B (Fig. 3b) exhibits only the 2Ϫ GR(CO3 ) bands: i.e., H2O and -OH stretching modes at 3500, Ϫ1 Ϫ1 ␦ 3410, 3310 cm , and 3170 cm ;H2O bending vibration ( ) Ϫ1 2Ϫ ␯ at 1630 cm ;CO3 stretching modes at 1350 ( 3); and lattice Fe-OH modes at 845, 770, 510, and 480 cmϪ1 (Taylor, 1980; Legrand et al., 2001b, Peulon et al., 2003). The lack of the 1070 and 955 cmϪ1 bands in spectrum (b) is consistent with the complete removal of the initial ferrous suspension. The FTIR spectrum of the exGR-Fe(III) product is given in Figure 3c. The 470 and 655 cmϪ1 bands are assigned to stretching Fe-O modes in octahedral Fe-O(OH) coordination (Markov et al., 1990). The broad band between 2300 and 3600 cmϪ1, attributed to OH stretching vibrations in water molecules and hydroxyl groups, Fig. 4. TEM image of the ferric product sampled at point C, is the sum of three contributions located at 3465, 3290, and exGR-Fe(III). Oxidation of carbonate green rust into ferric phases 3501

2Ϫ Table 1. Chemical analysis data of products sampled at point B, GR(CO3 ), and at point C, exGR-Fe(III). Comparisons of experimental data of exGR-Fe(III) with theoretical data of ferrihydrite and feroxyhyte.

Total Iron Content (% wt)

a Fe(II) (%) EDTA Hematisation CO2 Content (% wt)

2Ϫ Ϯ Ϯ Ϯ Ϯ GR(CO3 ) samples taken at point B 66 2 (4) 51.2 0.2 (3) 52.9 0.3 (3) 5.8 0.2 (3) Carbonate green rustb 67 52.8 6.9 ExGR-Fe(III) samples taken at point C 0 (1) 50.1 Ϯ 0.4 (5) 53.4 Ϯ 0.6 (3) 5.0 Ϯ 0.2 (4) Ferrihydritec 0 58.2 0 Feroxyhyted 0 62.9 0

In brackets, we give the number of samples that were analyzed. Each value was obtained from one solid sample synthesized according to the procedure given in Figure 1 (point B or C). An average value and the corresponding standard deviation were then calculated. a ϭ KMnO4 titration; Fe(II)% nFe(II)/n(total iron). b II III Theoretical values calculated by assuming the formula [Fe4 Fe2 (OH)12] [CO3,2H2O] given by Drissi et al. (1995). c Theoretical values calculated by assuming the formula Fe5HO8.4H2O given by Towe and Bradley (1967). d Theoretical values calculated by assuming the formula ␦-FeOOH given by Chukhrov et al. (1977).

2؊ 2Ϫ 3.2. Synthesis and Oxidation of GR(CO3 ) Suspension In 3.2.3. Influence of the rate of GR(CO3 ) oxidation 0.04 mol/L NaHCO Solution 3 Figure 7a reports the same experiment as Figure 5a, except that the stirring rate was increased from 300 to 700 rpm at point 3.2.1. E-t and pH-t transients B. The increase of the stirring rate induces an increase of the oxygen supply, and then a decrease of the time duration of the Experiments were done in solutions with lower NaHCO3 concentration to address compositions more closely simulating groundwater or aquifers. Figure 5a reports the E-t and pH-t curves recorded during the oxidation of an Fe(II) suspension in ϭ 0.04 mol/L NaHCO3 solution at initial pH 8.6. The step A-B corresponds to the oxidation of the initial Fe(II) suspension into 2Ϫ GR(CO3 ), as revealed by the dark green colour of the sus- pension at point B, and FTIR analysis. The average oxidation rate, 5.6 ␮moles FeII minϪ1, is very close to that determined in Figure 1. A pH decrease from 8.6 to 7.5 is observed during this 2Ϫ step. The oxidation of GR(CO3 ) into ferric product takes place from B to C. The shape of the E-t transient from B to C strongly differs from what has been recorded during the oxi- dation experiment reported in Figure 1. Moreover, the average oxidation rate, 5.8 ␮moles FeII minϪ1, is 2 to 3 times larger than that of Figure 1. The FTIR analysis of the solid product obtained at point C (Fig. 5b) shows the characteristic bands of lepidocrocite, ␥-FeOOH, at 745, 1020, 1155, 2800 and 3150 cmϪ1 (Schwertmann and Cornell, 1991; Weckler and Lu¨tz, 1998), suggesting that another oxidation mechanism occurs. A small quantity of goethite is also detected (see shoulders at ϳ900 and 800 cmϪ1). Moreover, the formation of ferrihydrite cannot reasonably be excluded.

2Ϫ 3.2.2. Influence of pH on GR(CO3 ) oxidation

Some experiments similar to the one described in Figure 5a were carried out at different starting pH values, and conse- quently at different pH values during the step from B to C. The ferric products formed at point C were identified by means of FTIR and XRD measurements. Figure 6 indicates the ferric products that were obtained as a function of the pH range recorded during the B-to-C step. For pHs lower than 8, the Fig. 5. (a) E-t and pH-t curves recorded during the oxidation by air formation of lepidocrocite is favoured. For pHs greater than Ϫ2 of 10 M Fe(II) suspension in 0.04 mol/L NaHCO3 solution, starting 8.9, only exGR-Fe(III) is formed. For pHs between 8 and 8.9, pH ϭ 8.6, T ϭ 25°C, stirring 300 rpm; (b) FTIR spectrum of the ferric a mixture of both lepidocrocite and exGR-Fe(III) is obtained. product obtained at point C, L ϭ lepidocrocite; G ϭ goethite. 3502 L. Legrand, L. Mazerolles, and A. Chausse´

2؊ 3.3. Acid Titration of GR(CO3 ) Suspension Figure 9 reports the results of acid titration of 0.04 mol/L

NaHCO3 solutions (50 mL) with (curve (a)) or without (curve 2Ϫ (b)) 1.67 mM GR(CO3 ) suspension. Curve (a) was obtained as follows: 1 mol/L NaOH was added to a 0.04 mol/L NaHCO3 Ϫ2 solution until pH reached 11.3. Then, 10 M FeCl2 was added and the pH decreased down to 10.4. The conversion of the 2Ϫ starting ferrous suspension (10 mM) into GR(CO3 ) suspen- sion (1.67 mM ϭ 10/6) was done by air oxidation (step A to B). At point B, the pH was 10.2. The cell was closed and placed under Ar bubbling and then, the acid titration of the solution ϩ 2Ϫ GR(CO3 ) suspension was performed. The blank curve (b) was obtained from the acid titration of a 0.04 mol/L NaHCO3 ϩ NaOH solution with initial pH of 10.2. At first, both curves 2Ϫ exhibit the neutralization of CO3 species in the 10-9pH range. A pH pseudoplateau at ϳ8.4, that should be attributed to

Fig. 6. Influence of pH on the nature of predominant ferric products ϭ obtained at point C in 0.04 mol/L NaHCO3 solution; T 25°C; stirring 300 rpm. E ϭ exGR-Fe(III); L ϭ lepidocrocite.

B-to-C step (26 min. in Fig. 7a against 57 min. in Fig. 5a). ExGR-Fe(III) is predominantly formed at the completion of oxidation (Fig. 7c); only a very small amount of lepidocrocite can be detected.

Another set of experiments in which H2O2 was used as a soluble, strongly oxidizing species, was done in 0.04 mol/L

NaHCO3 solutions with starting pHs in the 7 to 10 range. At point B, the cell was closed, placed under argon bubbling, and Ϫ3 0.25 mL of 1 mol/L H2O2 (i.e., 5 10 M) was added to 2Ϫ GR(CO3 ) suspension. Figure 7b gives typical E-t and pH-t transients. A sharp increase of potential in less than 1 min, along with a colour change from dark green to light brown is observed, indicating the oxidation completion. The ferric prod- uct is exGR-Fe(III), whatever the starting pH (FTIR spectrum 2Ϫ not shown). Finally, in an experiment where a dried GR(CO3 ) powder was left in contact with air, the final ferric product was also exGR-Fe(III).

3.2.4. Influence of blocking adsorbed species on the 2Ϫ oxidation of GR(CO3 ) Figure 8 reports the same experiment as in Figure 5a, except Ϫ2 that 10 MNa2HPO4 was added at point B. The addition of Ϫ HPO4 ions affects the shape of the B-to-C step, as compared to Figure 5a—a sharp increase of E followed by a potential plateau near 0.2 V is obtained. This shape is similar to that of Figure 1. Analysis of the ferric product formed at point C was carried out by FTIR. The spectrum (not shown) indicates that it is exGR-Fe(III) with phosphate species adsorbed at the surface (a spectrum resembling the one of Fig. 3d is obtained). Some similar experiments performed at different pH values in the 7 Fig. 7. (a) E-t and pH-t curves recorded during the oxidation by air to10 range also gave the same final ferric product. On the other Ϫ2 of 10 M Fe(II) suspension in 0.04 mol/L NaHCO3 solution, starting hand, the formation of lepidocrocite became inhibited. ExGR- pH ϭ 8.6, T ϭ 25°C. From A to B, stirring 300 rpm, beyond B, stirring Fe(III) with silicate species adsorbed at the surface was ob- 700 rpm; (b) E-t and pH-t curves recorded during the oxidation of 10Ϫ2 tained in solutions where Na SiO was added instead of M Fe(II) suspension in 0.04 mol/L NaHCO3 solution, starting pH 2 3 ϭ 8.6, T ϭ 25°C, stirring 300 rpm. From A to B, oxidation by air. At Na2HPO4. Note that the substitution of carbonate ions for point B, the cell was placed under argon bubbling and 5 10Ϫ3 MHO phosphate or silicate ions inside GR(CO2Ϫ) interlayers is not 2 2 3 were added (0.25 mL of 1 mol/L H2O2); (c) FTIR spectrum of the ferric considered (Benali et al., 2001). product, exGR-Fe(III), obtained at point C in curve (a). Oxidation of carbonate green rust into ferric phases 3503

Fig. 8. E -t and pH -t curves recorded during the oxidation by air of Ϫ2 10 M Fe(II) suspension in 0.04 mol/L NaHCO3 solution, starting pH ϭ ϭ Ϫ2 8.6, T 25°C, stirring 300 rpm. 10 MNa2HPO4 were added at point B.

2Ϫ the acid-induced dissolution of GR(CO3 ) suspension, is ob- served in curve (a). The volume of acid can be estimated to 4 mL, corresponding to an addition of 4.8 mol Hϩ with respect to 2Ϫ one mole GR(CO3 ). Note that the pH values recorded during the pseudoplateau are not strictly equilibrium values, mainly due to the too large rate of HCl addition, 0.05 mL minϪ1.

2؊ 3.4. Solid-State Redox Cycling of GR(CO3 )/exGR- Fe(III) System

2Ϫ GR(CO3 ) thin layers were electrochemically synthesized at the surface of an iron disk as described in part 2.2. Figure 10a shows the morphology of a layer after its oxidation by dis- Fig. 10. (a) SEM image of an exGR-Fe(III) thin layer on iron disk; (b) E-t transient recorded during the galvanostatic reduction (- 10 ␮A) Ϫ3 ϩ Ϫ2 of an exGR-Fe(III) thin layer in deaerated CaCO3 210 M 210 M NaCl solution at 25°C; (c) FTIR analysis of the disk surface before (c1) and after (c2) galvanostatic reduction. (c1) exGR-Fe(III); (c2) 2Ϫ GR(CO3 ); (*) traces of calcite.

ϩ solved oxygen in an aerated 0.01 mol/L NaHCO3 0.001 ϭ mol/L Na2HPO4 solution at pH 8.5. ExGR-Fe(III) is formed (see FTIR spectrum (c1) in Fig. 10c); no change of the layer morphology is observed, as compared to the starting 2Ϫ GR(CO3 ) particles. The hexagonal particles obtained from electrochemical synthesis have significantly larger sizes than those obtained from a suspension (compare Figs. 4 and 10a). Figure 10b reports the E-t transient obtained during the gal- vanostatic reduction of an exGR-Fe(III) layer in simulated groundwater. A plateau is observed, indicating the occurrence of the exGR-Fe(III) reduction process. The potential values recorded on the plateau are close to -0.4 V. The completion of exGR-Fe(III) reduction is revealed by the sharp decrease of the potential (time ϳ 4500 s), down to a range corresponding to the Fig. 9. 0.1 mol/L HCl acid-titration curve of 0.04 mol/L NaHCO3 2Ϫ electrolyte reduction. The FTIR analysis of the layer, per- solution containing (a) 1.67 mM of GR(CO3 ) suspension; (b) no 2Ϫ GR(CO3 ). formed after the galvanostatic reduction (Fig. 10c, spectrum 3504 L. Legrand, L. Mazerolles, and A. Chausse´

Table 2. Stability or conversion of exGR-Fe(III) under various wet aging conditions Polytetrafluoroethylene (PTFE).

Initial Solution Container pH T°C Storage Time Final

Ϫ2 ϩ Dried powder 10 M NaHCO3 Teflon 8.5 25 24 days ExGR-Fe(III) traces of goethite Ϫ2 Dried powder 10 M NaHCO3 Teflon 8.5 60 8 days Goethite Dried powder H2O Teflon 8.5 60 8 days Goethite ϩ Suspension left in solution 0.4 M NaHCO3 Teflon 9.6 60 8 days Hematite goethite after synthesis ϩ Suspension left in solution 0.4 M NaHCO3 Glass 9.6 50 40 days ExGR-Fe(III) detection of adsorbed after synthesis silicate ions Ϫ2 ϩ Ϫ3 ϩ Dried powder 10 M NaHCO3 10 M Teflon 8.5 60 8 days ExGR-Fe(III) adsorbed phosphate or Na2HPO4 or Na2SiO3 silicate ions

2؊ (c2)) shows the complete transformation of exGR-Fe(III) into 4.2. Solid-State Oxidation of GR(CO3 ) 2Ϫ GR(CO3 ), involving a solid-state reduction. 4.2.1. ExGR-Fe(III), a ferric oxyhydroxycarbonate

The exGR-Fe(III) product resulting from the solid-state ox- 3.5. Stability of exGr-Fe(III) In Solution 2Ϫ idation (SSO) of GR(CO3 ) exhibits a light brown colour and a hexagonal platelets morphology. Table 1 indicates that its ExGR-Fe(III) samples were stored in solutions for several Ϫ composition is very close to that of GR(CO2 ). From the days, and solids were then analysed by means of FTIR or XRD. 3 results of Table 1 and the charge balance (see Appendix for Results are computed in Table 2.At25°C, the conversion of details), the chemical composition of exGR-Fe(III) can be exGR-Fe(III) is very slow since only traces of goethite, proposed, (FeIII) (O) (H) (CO ) . The presence of re- ␣-FeOOH, are observed after 24 d aging. At 60°C, when the 6 14.41 12.30 3 0.74 maining carbonate ions inside the exGR-Fe(III) crystal lattice is suspensions are stored in a PTFE vessel (no silicate ions), the fully demonstrated in this study. Replacing O and H contents complete conversion into goethite or into a hematite (␣-Fe O ) Ϫ Ϫ 2 3 by O2 ,OH , and H O, the following formula can be pro- and goethite mixture occurs. When phosphate (or metasilicate) 2 posed, FeIII O (OH) (H O) (CO ) . Hence, ions are introduced at the beginning of the storage, or when a 6 (2.11ϩx) (12.30-2x) 2 x 3 0.74 ExGR-Fe(III) is a ferric oxyhydroxycarbonate. Finally, the glass container is used, exGR-Fe(III) does not transform due to Ϫ assumption that the solid-state oxidation of GR(CO2 ) gives the adsorption of phosphate or silicate species (see Fig. 3d) 3 ferrihydrite or feroxyhyte (Taylor, 1980; Hansen, 1989; Abdel- which inhibit the dissolution step needed for the formation of moula et al., 1996) must be revisited. goethite.

4.2.2. Deprotonation as a way for keeping electrical neutrality 4. DISCUSSION 2Ϫ The solid-state oxidation (SSO) of GR(CO3 ) involves the 2؊ 2ϩ 3ϩ 4.1. Carbonate Green Rust GR(CO3 ) transformation of Fe ions into Fe ions inside the crystal 2Ϫ lattice. The formula of GR(CO3 ) can be written as follows: II III The specific surface area of the Fe(II) suspension might be [Fe 4Fe 2(OH)12] [CO3, 2H2O]. This last formula and the Ϫ very large due to its amorphous state; this feature warrants an corresponding molar mass 635.1 g mol 1 can be admitted for II 2Ϫ easy dissolution and a flow of soluble Fe species large enough our GR(CO3 ). The solid-state oxidation implies charge com- to compensate the flow of dissolved oxygen. This fact is sup- pensation to satisfy the electrical neutrality of the whole struc- ported by the rather stable potential values that are recorded ture. The compensation by some anionic species during 2Ϫ along the A-to-B oxidation step. The oxidation of the soluble GR(CO3 ) oxidation can be ruled out since the number of FeII species supplied by the amorphous Fe(II) suspension might interlayer sites required for this purpose would be greater than lead to the production of soluble ferrous-ferric intermediate the available sites (Legrand et al., 2003). The loss of protons to species, whose occurrence has previously been claimed by accommodate the charge compensation is more consistent with Misawa et al. (1973 and 1974). According to these authors, the the results of Table 1. The solid-state oxidation (SSO) of II III 2Ϫ following schematic formula may be proposed [(Fe )2(Fe )1]. GR(CO3 ) can be written as: 2Ϫ The precipitation of GR(CO3 ) occurs as the concentration of II III III II III ͓Fe Fe (OH) ][CO ,2HO] 3 Fe O ϩ (OH) Ϫ [(Fe )2(Fe )1] species exceeds the solubility limit. From Table 4 2 12 3 2 6 (2 x) (12 2x) 2Ϫ 1, the following formula for GR(CO3 ) can be proposed (see ϫ ͑H O) (CO ͒ ϩ 4Hϩ ϩ 4eϪ (3) Appendix for details), 2 x 3 The value of x depends on whether or not the deprotonation of II III Ϫ Fe3.96Fe2.04(OH)12.34(CO3)0.85(H2O)2.62 . (2) interlayer water molecules to give OH ions is considered. Eqn. 3 shows that exGR-Fe(III) is obtained from the release of 2Ϫ The formula is consistent with those reported by Hansen 4 protons and 4 electrons by GR(CO3 ). Combining Eqn. 3 (1989), Stampfl (1969) and Drissi et al. (1995), with the reduction reaction of O2 or H2O2, reactions (4) and (5) II III [Fe 4Fe 2(OH)12] [CO3, mH2O], with m likely equal to (2). can be established: Oxidation of carbonate green rust into ferric phases 3505

͓ II III ϩ 3 III Fe4 Fe2 (OH)12][CO3,2H2O] O2 Fe6 O(2ϩx)(OH)(12Ϫ2x) ϫ ͑ ϩ H2O)x(CO3) 2H2O (4)

͓ II III ϩ 3 III Fe4 Fe2 (OH)12][CO3,2H2O] 2H2O2 Fe6 O(2ϩx) ϫ ͑ ϩ OH)(12Ϫ2x)(H2O)x(CO3) 4H2O . (5)

4.3. Stability of the Ferric Oxyhydroxycarbonate, exGr- Fe(III)

At ambient temperature and below, exGR-Fe(III) remains stable even after large periods of aging, either as a dry powder or as particles in solution. The conversion into thermodynam- ically stable ferric phases, goethite or hematite, occurs only in moderately heated solutions stored in glass-free containers. The formation of the latter compounds requires the occurrence of a dissolution-precipitation (D-P) mechanism (Jambor and Dutri- zac, 1998). In presence of silicate or phosphate species, the dissolution step is inhibited due to the adsorption of these species onto the exGR-Fe(III) surface (Sigg and Stumm, 1981; Cornell et al., 1987), and the latter compound transformation is prevented. Fig. 11. Schematic representation of formation and transformation 2؊ 2Ϫ 4.4. Oxidation of GR(CO3 ) Via Solution pathways of GR(CO3 ).

2Ϫ At point B, the solution contains GR(CO3 ) suspension and II III soluble [(Fe )2(Fe )1] species (Misawa et al., 1973; Misawa et should depend on the experimental conditions; lepidocrocite al., 1974) in equilibrium with the solid phase, as illustrated by predominantly formed in the present study, goethite was ob- the following schematic reaction, tained in synthesis from Drissi et al. (1995), Schwertmann and dissolution Cornell (1991), ferrihydrite may be present along with lepido- ͓ II III |- 0 ͓͑ II III Fe4 Fe2 (OH)12][CO3,2H2O] 2 Fe )2(Fe )1]. crocite and goethite. The DOP mechanism may be extended to precipitation sulphate and chloride green rusts as the occurrence of ferrous- ferric species (green complexes) has also been reported in (6) neutral and mildly alkaline solutions containing sulphate or 2Ϫ Figure 9 showed that the dissolution of GR(CO3 ) is promoted chloride ions (Misawa et al., 1973; Misawa et al., 1974). by acid addition and that the pH range where it occurs, ϳ8.4, is less alkaline than the one relative to the neutralisation of 4.5. Formation and Transformation Pathways of 2Ϫ 2 CO3 species according to GR(CO3) 2Ϫ ϩ ϩ 3 Ϫ CO3 H HCO3 . (7) Results of the present study allow the establishment of a

Ϫ schematic representation for formation and transformation 2 Ϫ Ϫ On the other hand, protons will preferentially react with CO3 2 2 2Ϫ pathways of GR(CO3 )(Fig. 11). The formation of GR(CO3 ) (if available) rather than with GR(CO3 ). from amorphous Fe(II) suspension involves three steps, as The introduction of oxygen into the cell at point B induces II III proposed in section 4.1: dissolution of amorphous Fe(II) sus- the oxidation of [(Fe )2(Fe )1] species according to the sche- pension, oxidation of soluble ferrous species into soluble fer- matic reaction 2Ϫ rous-ferric species, and precipitation to GR(CO3 ). The oxi- 2Ϫ ϩ 2͓͑FeII) (FeIII) ] ϩ O 3 6 FeOOH ϩ yH . (8) dation of GR(CO3 ) can proceed by solid-state reaction (SSO) 2 1 2 or by reaction via solution (DOP). The SSO mechanism is The production of protons associated to the oxidation of FeII- promoted at the expenses of the DOP mechanism (i) by in- 2Ϫ 2Ϫ containing species into FeOOH is well established (Schwert- creasing the [CO3 ]/[GR(CO3 )] ratio, (ii) by increasing the mann and Cornell, 1991). In 0.04 mol/L NaHCO3 solution with oxidation rate, and (iii) in the presence of adsorbed blocking 2Ϫ very slightly alkaline pH, CO3 concentration in solution is species such as phosphate or silicate. Condition (i) is monitored very low and reaction (7) is negligible. Then, the protons by adjusting the total concentration of carbonate species, the produced in Eqn. 8 can promote reaction (6) in the dissolution pH, and the concentration of iron species. Condition (ii) is II III sense, leading to the release of fresh [(Fe )2(Fe )1] species in related to the flow of oxygen (solution stirring) or soluble solution. A self-sustained mechanism is observed where oxidising species (addition rate). Condition (iii) may be ful- 2Ϫ GR(CO3 ) and dissolved oxygen are the reagents, FeOOH the filled for many soil or corrosion environments since phosphate II III 2Ϫ end product, and [(Fe )2(Fe )1] the intermediate species. The and silicate ions are widespread species. The GR(CO3 ) oxi- whole transformation is a dissolution-oxidation-precipitation dation mechanism (see Introduction) proposed by Benali et al. (DOP) mechanism. The nature of the ferric oxihydroxide (2001) is not consistent with the results of the present paper. In 3506 L. Legrand, L. Mazerolles, and A. Chausse´ our mind, their mechanism is quite doubtful since it is well accompanying the dissimilatory reduction of hydrous ferric oxide by known that the ferrihydrite-to-goethite transformation is a very a groundwater bacterium. Geochim. Cosmochim. Acta 62, 3239– slow reaction at ambient temperature (Schwertmann and Cor- 3257. Ge´hin A., Ruby C., Abdelmoula M., Benali O., Ghanbaja J., Refait Ph., nell, 1991; Jambor and Dutrizac, 1998). As a matter of fact, the and Ge´nin J. M.-R. (2002) Synthesis of Fe(II-III) hydroxysulphate ferric product they synthesized might be exGR-Fe(III) rather green rust by coprecipitation. Solid State Sci. 4, 61–66. than ferrihydrite. The solid-state reduction of exGR-Fe(III) Ge´nin J. M. R., Olowe A. A., Re´siak B., Benbouzid-Rollet N. D., 2Ϫ Confente M. and Prieur D. (1993) Identification of sulphated green back to GR(CO3 ) was demonstrated in our present study. This reduction process should occur with a negligible release of rust 2 compound produced as a result of microbially induced corro- sion of steel sheet piles in harbour. In Marine Corrosion of Stainless soluble iron species in solution, as for the oxidation process. If Steels: Chlorination and Microbial Effects, European Federation it occurs in natural environments, the solid-state redox cycling Corrosion Series n¡10, The Institute of Materials, London, 10, 162– 2Ϫ of GR(CO3 ) may affect the remobilization of iron. To state, 166. the reaction pathways described in the present study should be Ge´nin J. M. R., Olowe A. A., Refait Ph., and Simon L. (1996) On the considered in all environments where GR(CO2Ϫ) occurs. stoichiometry and Pourbaix diagram of Fe(II)-Fe(III) hydroxy-sul- 3 phate or sulphate-containing green rust 2: an electrochemical and Mo¨ssbauer spectroscopy study. Corros. Sci. 38, 1751–1762. Acknowledgments—The authors thank the associate-editor, G. Sposito, Ge´nin J. M. R., Refait Ph., Simon L., and Drissi S. H. (1998) Prepa- and two of the reviewers for their helpful and valuable comments and ration and Eh-pH diagrams of Fe(II)-Fe(III) green rust compounds; suggestions. hyperfine interaction characteristics and stoichiometry of hydroxy- chloride, -sulphate and –carbonate. Hyp. Interact. 111, 313–318. Associate editor: G. Sposito Hansen H. C. B. (1989) Composition, stabilization, and light absorp- tion of Fe(II)Fe(III) hydroxy-carbonate (“green rust”). Clay Miner. 24, 663–669. REFERENCES Hansen H. C. B. and Bender Koch C. (1998) Reduction of nitrate to ammonium by sulphate green rust: activation energy and reaction Abdelmoula M., Refait Ph., Drissi S. H., Mihe J. P., and Ge´nin J. M.-R. mechanism. Clay Minerals 33, 87–101. (1996) Conversion electron Mo¨ssbauer spectroscopy and X-ray dif- Jambor J. L. and Dutrizac J. E. (1998) Occurrence and constitution of fraction studies of the formation of carbonate-containing green rust natural and synthetic ferrihydrite, a widespread iron oxyhydroxide. ϩ one by corrosion of metallic iron in NaHCO3 and (NaHCO3 NaCl) Chem. Rev. 98, 2549–2585. solutions. Corros. Sci. 38, 623–633. Legrand L., Savoye S., Chausse´ A., and Messina R. (2000) Study of Allmann R. (1968) The crystal structure of pyroaurite. Acta Cryst. B24, oxidation products formed on iron in solutions containing bicarbon- 972–977. ate/carbonate. Electrochim. Acta 46, 111–117. Benali O., Abdelmoula M., Refait Ph., and Ge´nin J. M.-R. (2001) Legrand L., Abdelmoula M., Ge´hin A., Chausse´ A., and Ge´nin J. Effect of orthophosphate on the oxidation product of Fe(II)-Fe(III) -M. R. (2001a) Electrochemical formation of a new Fe(II)-Fe(III) hydroxycarbonate. The transformation of green rust to ferrihydrite. hydroxy-carbonate green rust: characterisation and morphology. Geochim. Cosmochim. Acta 65, 1715–1726. Electrochim. Acta 46, 1815–1822. Bernal J. D., Dasgupta D. R., and Mackay A. L. (1959) The oxides and Legrand L., Sagon G., Lecomte S., Chausse´ A., and Messina R. of iron and their structural inter-relationships. Clay (2001b) A. Raman and infrared study of a new carbonate green rust Miner. Bull. 4, 15–30. obtained by electrochemical way. Corros. Sci. 43, 1739–1749. Bonin P. M., Jedral W., Odziemkowski M. S., and Gillham R. W. Legrand L., Maksoub R., Sagon G., Lecomte S., Dallas J. P., and (2000) Electrochemical and Raman spectroscopic studies of the Chausse´ A. (2003) Electroanalytical and kinetic investigations on the influence of chlorinated solvents on the corrosion behaviour of iron carbonate green rust –Fe(III) redox system. J. Electrochem. Soc. 150, in borate buffer and in simulated groundwater. Corros. Sci. 42, B45–B51. 1921–1939. Butler G. and Beynon J. G. (1967) The corrosion of mild steel in Lindsay W. L. (1979) Chemical Equilibria In Soils. Wiley Interscience. boiling salt solutions. Corros. Sci. 7, 385–404. Loyaux-Lawniczak S., Refait Ph., Ehrhardt J. J., Lecomte P., and Ge´nin Cornell R. M., Giovanoli R., and Schindler P. W. (1987) Effect of J. M. R. (2000) Trapping of Cr by formation of ferrihydrite during silicate species on the transformation of ferrihydrite into goethite and the reduction of chromate ions by Fe(II)-Fe(III) hydroxysalt green hematite in alkaline media. Clays Clay Miner. 35, 21–28. rusts. Environ. Sci. Technol. 34, 438–443. Chukhrov F. V., Zvyagin B. B., Gorshkov A. I., Ermilova L. P., McGill I. R., McEnaney B., and Smith D. C. (1976) Crystal structure Korovushkin V. V., Rudnitskaya Y. E., and Yakubovskaya N. Y. of green rust formed by corrosion of cast iron. Nature 259, 1521– (1977) Feroxyhyte, a new modification of FeOOH. Int. Geol. Rev. 1529. 19, 873–890. Markov L., Blaskov V., Klissurski D., and Nikolov S. (1990) The Detournay J., Derie´ R., and Ghodsie M. (1975) The region of stability thermal decomposition mechanism of Iron(III) hydroxide carbonate ␣ of GR II in the electrochemical potential-pH diagram in sulphate to -Fe2O3. J. Mater. Sci. 25, 3096–3100. medium. Corros. Sci. 15, 295–306. Misawa T., Hashimoto K., and Shimodaira S. (1973) Formation of Detournay J., De´rie´ R., and Ghodsie M. (1976) Etude de l’oxydation Fe(II)1-Fe(III)1 intermediate green complex on oxidation of ferrous in neutral and slightly alkaline sulphate solutions. J. Inorg. Nucl. par ae´ration de Fe(OH)2. Z. Anorg. Allg. Chem. 427, 265–273. Drissi S. H., Refait Ph., Abdelmoula M., and Ge´nin J. M. R. (1995) The Chem. 35, 4167–4174. preparation and thermodynamic properties of Fe(II)-Fe(III) hydrox- Misawa T., Hashimoto K., and Shimodaira S. (1974) The mechanism ide-carbonate (green rust 1); Pourbaix diagram of iron in carbonate- of formation of and oxyhydroxides in aqueous solutions at containing aqueous media. Corros. Sci. 37, 2025–2041. room temperature. Corros. Sci. 14, 131–149. Erbs M., Hansen H. C. B., and Olsen C. E. (1999) Reductive dechlo- Olowe A. A. and Ge´nin J. M. R. (1991) The mechanism of oxidation rination of carbon tetrachloride using iron(II) iron(III) hydroxide of Fe(II) hydroxide in sulphated aqueous media: importance of the (green rust). Environ. Sci. Technol. 33, 307–311. initial ratio of the reactants. Corros. Sci. 32, 965–984. Erdo¨s V. E. and Altorfer H. (1976) Ein dem Malachit a¨hnliches Ona-Nguema G., Abdelmoula M., Jorand F., Benali O., Ge´hin A., basisches Eisenkarbonat als Korrosiosprodukt von Stahl. Werkstoffe Block J. C., and Ge´nin J. M. R. (2002) Iron (II, III) hydroxycarbon- und Korrosion 27, 304–312. ate green rust formation and stabilization from lepidocrocite biore- Feitknecht W. and Keller G. (1950) U¨ ber die Dunkelgru¨nen Hydroxy- duction. Environ. Sci. Technol. 36, 16–20. verbindungen des eisens. Z. Anorg. Allg. Chem. 262, 61–68. Peulon S., Legrand L., Antony H. and Chausse´ A. (2003) Electrochem- Fredrickson J. K., Zachara J. M., Kennedy D. W., Dong H., Onstott ical deposition of thin films of green rusts 1 and 2 on inert gold T. C., Hinman N., and Li S. M. (1998) Biogenic iron mineralization substrate. Electrochem. Comm. 5, 208–213. Oxidation of carbonate green rust into ferric phases 3507

Ponnamperuma F. N. (1972) The chemistry of submerged soils. Adv. In APPENDIX Agronomy 24, 173–189. Refait Ph. and Ge´nin J. M. R. (1993) The oxidation of Fe(II) hydroxide Chemical Composition of Carbonate Green Rust GR(CO2) in chloride-containing aqueous media and Pourbaix diagrams of 3 green rust I. Corros. Sci. 34, 797–819. 2Ϫ FeII content 66%; average total iron content (FeII ϩ FeIII) 52.1% wt; Refait Ph., Simon L., and Ge´nin J. M. R. (2000) Reduction of SeO4 anions and anoxic formation of iron(II)-iron(III) hydroxy-selenate CO3 content 7.91% wt. green rust. Environ. Sci. Technol. 34, 819–825. Calculations for 100 g: Savoye S., Legrand L., Sagon G., Lecomte S., Chausse´ A., Messina R., II ϭ III ϭ ϭ and Toulhoat P. (2001) Experimental investigations on iron corro- n(Fe ) 0.616 mole; n(Fe ) 0.317 mole; n(CO3) 0.132 mole sion products formed in bicarbonate/carbonate-containing solutions Charge balance and mass balance at 90°C. Corros. Sci. 43, 2049–2064. Schwertmann U. and Cornell R. M. (1991) Iron Oxides In the Labo- n͑OH)ϭ2 n(FeII) ϩ 3 n(FeIII) Ϫ 2 n(CO ) ϭ 1.919 moles ratory: Preparation and Characterization. VCH Publishers, Inc.. 3 Schwertmann U. and Fechter H. (1994) The formation of green rust and n͑H O͒ ϭ ͓100 Ϫ 55.85 n͑FeII) Ϫ 55.85 n͑FeIII) Ϫ 60 n͑CO ͒ its transformation to lepidocrocite. Clay Miner. 29, 87–92. 2 3 Sigg L. and Stumm W. (1981) The interactions of anions and weak Ϫ 17 n͑OH͔͒/18 ϭ 0.408 mole acids with the hydrous goethite (␣-FeOOH) surface. Collo¬õd. Sur- faces 2, 101–117. The chemical composition is then determined for 6 Fe atoms Stampfl P. P. (1969) Ein basisches Eisen II-III Karbonat in Rost. Corros. Sci. 9, 185–187. Chemical Composition of Ferric Oxyhydroxycarbonate ExGR- Taylor R. M. (1980) Formation and properties of Fe(II)-Fe(III) hy- Fe(III) droxycarbonate and its possible significance in soil formation. Clay

Miner. 15, 369–382. II III Towe K. M. and Bradley W. F. (1967) Mineralogical constitution of Fe content 0%; average total iron content (Fe ) 51.8% wt; CO3 collo¨ıdal “hydrous ferric oxide.” J. Collo¬õd. Interface Sci. 24, 384– content 6.82% wt. 392. Calculations for 100 g: Trolard F., Ge´nin J. -M. R., Abdelmoula M., Bourrie´ G., Humbert B., n͑FeIII) ϭ 0.927 mole ; n͑CO ͒ ϭ 0.114 mole and Herbillon A. (1997) Identification of a green rust mineral in a 3 reductomorphic soil by Mo¨ssbauer and Raman spectroscopies. Charge balance and mass balance Geochim. Cosmochim. Acta 61, 1107–1111. Weckler B. and Lu¨tz H. D. (1998) Lattice vibration spectra. Part XCV. ͑ III Ϫ ͑ ͒ Ϫ ͑ ͒ ϩ ͑ ͒ ϭ 3nFe ) 2n CO3 2nO n H 0 Infrared spectroscopic studies on the iron oxide hydroxides goethite (␣), akagane´ite (␤), lepidocrocite (␥), and feroxyhite (␦). Eur. J. ͑ III ϩ ͑ ͒ ϩ ͑ ͒ ϩ ͑ ͒ ϭ 55.85 n Fe ) 60 n CO3 16 n O 1nH 100 Solid State Inorg. Chem. 35, 531–544. Williams A. G. B. and Scherer M. M. (2001) Kinetics of Cr(VI) n͑O͒ ϭ 2.227 mole; n͑H͒ ϭ 1.900 reduction by carbonate green rust. Environ. Sci. Technol. 35, 3488– 3494. The chemical composition is then determined for 6 Fe atoms.